Use cases
Every use case follows the same steps: create an endpoint, connect a source, then build dashboards and alerts. The table points you to the right guide.
| Use case | What you learn | Start here |
|---|---|---|
| Website traffic | Visitors, sources, engagement, Web Vitals | Website analytics |
| Product behavior | How signed-in users use features, funnels, drop-off | Product analytics |
| AI agent debugging | What each run did, which step failed or was slow | Debug AI agents |
| Production AI quality | Errors, latency, cost and evaluation scores over time, with alerts | Monitor in production |
| Coding-tool spend | Claude Code, Codex and Antigravity usage and cost per developer and project | Track AI and coding spend |
| AI to business outcomes | How AI behavior affects conversions, tickets or revenue | Product analytics |
| Network and infrastructure | Device and server health, security signals | Network and infrastructure |
| IoT and devices | Sensor readings, heartbeats, fleet status | Send any data |
| Business processes | Where orders, claims or onboarding get stuck | Business process monitoring |
Website traffic and performance
Add the tracker once to your shared footer and get page views, sessions, acquisition channels, engagement, scroll depth, clicks and Core Web Vitals. Send server-side events to the same workspace to see backend latency next to user behavior.
Tutorial: How to Track Website Traffic and Performance with Anosys
AI agents, from development to production
Use the SDK for your framework (OpenAI, Anthropic, Claude Agent SDK, or any OpenTelemetry source) to trace every run. In development, open single runs in the Trace Timeline. In production, add a dashboard, alerts and scheduled evaluations.
Background reading: What Is AI Observability, and Why Current Tools Are Failing You
Coding-tool usage and spend
Install the hook for Claude Code, Codex CLI or Antigravity on each developer machine. See tokens, estimated cost, tool calls and files changed per developer and project, with optional content redaction.
Connecting AI to business outcomes
Send the same user or process ID from your website tracker, your backend events and your AI SDK. Then one dashboard can show, for example, whether users who got a slow or low-scoring AI answer converted less.
Business processes
Send one event per workflow step with a shared process ID. Chart the flow with a Funnel Plot or Sunkey Diagram, compute durations with process units, and alert on stuck items.
